Agrammatic patterns in Alzheimer's disease: Evidence from tense, agreement, and aspect

Abstract

Background: Little research has been conducted on functional categories in probable Alzheimer's disease (pAD). Furthermore, the findings are contradictory, since some studies report ceiling performance on tense and/or subject–verb agreement (Kaprinis & Stavrakaki, 2007; Kavé & Levy, 2003), whereas others report morphosyntactic deficits and agrammatic profiles (e.g., Altmann, Kempler, & Andersen, 2001). Aims: This study investigates the ability of Greek-speaking pAD individuals to produce and judge subject–verb agreement, tense, and aspect. Given pAD individuals have working memory limitations (e.g., Baddeley, 1996), and given the differential processing demands of agreement, tense and aspect (e.g., Fyndanis, Varlokosta, & Tsapkini, 2012a), pAD participants are expected to perform better on agreement than on tense/aspect. Based on the hypothesis that reference to the past is computationally more demanding than reference to the future/present (e.g., Bastiaanse et al., 2011), a within-tense dissociation is expected to emerge. Further, on the assumption that unmarked values of functional categories are less demanding than marked values (e.g., Lapointe, 1985), the imperfective (unmarked) aspect is expected to be better preserved than the perfective (marked) aspect. Methods & Procedures: Ten Greek-speaking mild pAD individuals and six healthy controls participated in a sentence completion task, a grammaticality judgement task, and a sentence–picture matching task. Non-parametric tests were used for analysis of results. Outcomes & Results: PAD participants were found to be significantly more impaired in aspect compared to tense and agreement, in both production and grammaticality judgement/comprehension. Agreement was found significantly better preserved than tense in production. Similar patterns of performance have been attested in agrammatism (e.g., Fyndanis et al., 2012a). Reference to the past and reference to the future did not dissociate, whereas the imperfective aspect was found to be significantly more impaired than the perfective aspect in production. Conclusions: PAD participants' better performance on producing agreement, compared to tense/aspect, is accounted for in terms of the differential demands these categories pose on the processing system. Agreement is computationally less demanding than tense/aspect, because the former involves processing of grammatical information only, whereas the latter involve processing and integration of grammatical and extralinguistic/conceptual information. The preponderance of tense over aspect is attributed to the subjectivity of the latter, which renders it either a category “difficult” to test, or a computationally more demanding condition. The results also show that reference to the past is as demanding as reference to the future. The hypothesis that unmarked values are easier than marked ones is not supported by our data.